Social capital and depressive symptoms: The association of psychosocial and network dimensions of social capital with depressive symptoms in Montreal, Canada

Social Science & Medicine 86 (2013) 96e102

Contents lists available at SciVerse ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Social capital and depressive symptoms: The association of psychosocial and network dimensions of social capital with depressive symptoms in Montreal, Canada Emma Bassett a, Spencer Moore a, b, * a b

School of Kinesiology and Health Studies, Queen’s University, Canada Department of Community Health and Epidemiology, Queen’s University, Canada

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 19 March 2013

Depression is the most common mental illness worldwide, and although aspects of the social environment, including social capital, have been linked to depression, the underlying mechanisms are not well understood. In this study, we assessed whether (1) network and psychosocial dimensions of individual social capital were each associated with depressive symptoms, and (2) the association varied according to the location of the capital, i.e., outside or inside a person’s neighbourhood. The current study used data from the Montreal Neighbourhood Networks and Healthy Ageing Study (MoNNET-HA). MoNNET-HA consisted of a representative sample of 2707 adults from 300 census tracts in the Montreal Metropolitan Area. The CESD-10 instrument was used to assess the presence of depressive symptoms with a cut off of more than three symptoms used to indicate depressive symptomatology. Name and position generator instruments were used to assess the existence of a core tie, core tie diversity, and network social capital both inside and outside the neighbourhood. Questions on generalized trust, trust in neighbours, and neighbourhood cohesion were used to assess psychosocial dimensions of social capital inside and outside the neighbourhood. Community and general group participation were also included as structural dimensions of social capital. Analyses adjusted for a range of socio-demographic and economic characteristics. Results from multilevel logistic regressions indicated that the core tie diversity as well as the psychosocial dimensions of generalized trust, trust in neighbours, and perceptions of neighbourhood cohesion reduced the likelihood of depressive symptoms in urban-dwelling adults. Network and psychosocial components of social capital within neighbourhood contexts should be considered when examining social capital and depressive symptoms. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: Canada Social capital Social networks Depressive symptoms Adults Urban health

Introduction Major depressive disorder is the most common mental illness worldwide, affecting 3.2% of the world’s population, and includes symptoms such as low mood and loss of interest in daily activities (American Psychiatric Association, 2000; Moussavi et al., 2007). It has been projected that by 2020, depression will rank as the second leading cause of disease burden worldwide (Moussavi et al., 2007). Ever since the work of Emile Durkheim, researchers have acknowledged and sought to examine the influence of the social environment and social relationships on mental health and

* Corresponding author. 28 Division Street, School of Kinesiology and Health Studies, Kingston, ON K7L 3N6, Canada. Tel.: þ1 613 533 6000; fax: þ1 613 533 2009. E-mail address: [email protected] (S. Moore). 0277-9536/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.socscimed.2013.03.005

depression. Social capital is a relatively more recent construct in the social epidemiological research field, and generally refers to the resources which individuals and, potentially, groups have access to through their social connections (Bourdieu, 1986). Social capital is often considered to be a feature of one’s social environment and to have psychosocial, structural, and network components. Despite recent advances in research on social capital and mental health, less is known about the contribution of network versus psychosocial mechanisms in the link between social capital and depression. To develop a greater understanding of those links, more research needs to investigate the relative associations between psychosocial and network components of social capital and depression. A person may be able to draw on their social capital within various contexts, including workplace, school, and neighbourhood. Assessing the contextual sources of a person’s social capital, specifically, in this case, inside and outside one’s neighbourhood, may be important for research and health promotion purposes (Moore

E. Bassett, S. Moore / Social Science & Medicine 86 (2013) 96e102

et al., 2011). For example, outside-neighbourhood social capital may be more beneficial than inside-neighbourhood social capital for a person’s self-rated health (Moore et al., 2011). Outside ties may indicate greater heterogeneity in social connections and access to a greater diversity of resources (Moore et al., 2011). Yet, few studies have differentiated between general- and neighbourhoodspecific social capital when examining social capital and depression. Results from such findings might better guide the content and types of treatment and prevention programs targeting depression.

97

participation components of social capital to be associated with better self-reported mental health. Moore et al. (2011) showed that outside neighbourhood network diversity and generalized trust were associated with self-reported health. In a study on physical inactivity, network diversity but not generalized trust was shown to decrease the likelihood of physical inactivity (Legh-Jones & Moore, 2012). Findings from such studies can help illuminate more precisely the potential mechanisms by which social capital influences health, and provide the type of information needed to guide social capital intervention planning.

Components of social capital Social capital and depression Within public health research, the concept of social capital is often operationalized using measures of trust, community participation, and individual- and community-level networks (Whitley & McKenzie, 2005). Differences in the operationalization of social capital have led to the rise of two main approaches e psychosocial and network e to understand the link between social capital and health. The psychosocial approach include constructs that are both social and psychological in nature and typically include measures of trust, norms, reciprocity, and perceptions of surrounding social environments (Legh-Jones & Moore, 2012). Although not strictly considered to be a psychosocial measure, community participation is often considered alongside trust as another dimension of social capital (Fujiwara & Kawachi, 2008; Veenstra, 2005). Strong associations have been found between the psychosocial components of individual social capital and various health outcomes, including depression (Aslund, Starrin, & Nilsson, 2010; De Silva, McKenzie, Harpham, & Huttly, 2005; Fujiwara & Kawachi, 2008; Sund, Jorgensen, Jones, Krokstad, & Heggdal, 2007; Veenstra, 2005; Webber, Huxley, & Harris, 2010). The psychosocial approach has however been criticized for measuring indirect or proxy measures of social capital, as well as the concepts that more closely relate to social cohesion than social capital (Carpiano, 2006; Lin, 1999). The network approach argues that social capital measures should directly assess how resources are accessed within social networks for personal benefit (Carpiano, 2006). Rather than focussing on measures of trust, network approaches measure directly how and to whom individuals are connected within their social structures, and how those connections give people access to a range of socially-valuable resources (Lin, 1999). Position generators have been used to assess a person’s access to social capital. Position generators include a list of various occupations, ranked according to their level of prestige (Lin, 1999). An individual with high levels of network capital would be someone who could reach higher ranked positions but who also has access to a diverse range of occupational positions (Lin, 1999). Although there can be downsides to social capital (Moore, Daniel, Gauvin, & Dubé, 2009; Moore, Daniel, Paquet, Dubé, & Gauvin, 2009), network capital may bring health benefits through the expressive and instrumental resources that people may access and mobilize via their social connections (Lin, 2001). Given the additional costs and research time often required to collect network data, few studies have been able to include both psychosocial and formal network measures when examining the relationship between social capital and health. Whereas most researchers tend to focus on one approach, the inclusion of both types of measures in studies of social capital can provide insight into which are most strongly associated with particular health outcomes, and which mechanisms might more closely link social capital to those outcomes. Recent studies that have included both types of measures have shown psychosocial, participation, and network measures to have different associations with health depending on the specific outcome being examined. For example, Carpiano and Hystad (2011) showed network, psychosocial, and

Research on individual social capital and depression has tended to rely on psychosocial measures of social capital, such as trust, and structural indicators of community participation. Several such studies have shown generalized trust at the individual level to be inversely related to depressive symptoms (Aslund et al., 2010; De Silva et al., 2005; Fujiwara & Kawachi, 2008; Sund et al., 2007; Veenstra, 2005; Webber et al., 2010). Particularized trust, e.g., trust in neighbours, has been shown to be a protective against depression (Fujiwara & Kawachi, 2008; Webber et al., 2010). Indicators of individual participation in community activities, however was not associated with depression (Fujiwara & Kawachi, 2008). Less research has examined individual social capital and depression using formal network measures of social capital. Using a resource generator to measure the social capital, Webber et al. (2010) did not find any association of social capital with depression over a six-month period. Haines, Beggs, and Hurlbert (2011) examined the association among neighbourhood disadvantage, network social capital and depressive symptoms, and found network capital to mediate the association between neighbourhood disadvantage and depressive symptoms (Haines et al., 2011). Research on social networks and depression has also highlighted the importance of social relationships for depressive symptoms. For example, studies investigating structural components of individual’s networks have found that people located centrally within social networks are less likely to be depressed than those on the periphery (Rosenquist, Fowler, & Christakis, 2011). Individuals are more likely to have depression if their direct (i.e., friends) and indirect social ties (i.e., friends of friends) also report depressive symptoms (Rosenquist et al., 2011). To gain a better understanding of the potential network mechanisms by which social capital might be associated with depression, more research using formal network measures of social capital are needed. There may also be a need to understand differences between the potential influences of a person’s overall social networks from those specific to the person’s neighbourhood. Identifying whether the source of person’s social capital is associated with depressive symptoms may aid in the design of health promotion programs that serve to target or not target the places where people reside. Purpose The current study aims to better understand the association of depressive symptoms with social capital using network, psychosocial, and participation measures of social capital. Adjusting for each type of measure in our models will enable the study to identify more clearly which mechanism may be more strongly associated with depressive symptoms. In addition, this study assesses whether the association between social capital and depressive symptoms differs depending on a person’s general or neighbourhood-specific social capital. This leads us to two research questions. First, are network, structural, or psychosocial dimensions of social capital more strongly associated with depressive symptoms when

98

E. Bassett, S. Moore / Social Science & Medicine 86 (2013) 96e102

controlling for socio-demographic variables? Second, does the geographical source (i.e., generally or inside one’s neighbourhood) of a person’s social capital matter in terms of the association between social capital and depressive symptoms? Findings from this research will contribute to a clearer understanding of the potential mechanisms linking social capital to mental health, and betterdefined social capital interventions seeking to address mental health issues within urban environments.

inside or outside the neighbourhood) with depressive symptoms, social capital measures were differentiated according to whether people’s connections or perceptions were specific to the neighbourhood or more general in orientation. Four network social capital variables were measured: (1) core social ties, (2) core tie diversity (3) general network capital, and (4) neighbourhood network capital. Whereas (1) and (3) measured general social ties and capital i.e., regardless of geographical source, numbers (2) and (4) were specific to individuals’ neighbourhoods.

Methods Study design We used data from the 2008 Montreal Neighbourhood Networks and Healthy Ageing Study (MoNNET-HA). Montreal Metropolitan Area (MMA) census tracts (N ¼ 862) were stratified into tertiles of high, medium, and low SES areas using median household income data from the 2001 Canada Census. Of the 862 census tracts in the MMA, one hundred were selected from each tertile for a total of 300 census tracts. To collect data from a range of age groups, respondents were stratified into three age categories: (1) 25e44 years old, (2) 45e64 years old, and (3) 65 years and older. With the exception of seven census tracts, three respondents were randomly selected from each age category within each census tract for a total of nine respondents per census tract (N ¼ 2679). In the seven remaining tracts, four respondents were selected due to the random polling procedure (N ¼ 28). Selection criteria required that respondents 1) had lived at their current address for at least 12 months, 2) were not currently institutionalized, and 3) were able to complete the questionnaire in either English or French. Random digit dialling of listed telephone numbers was used to contact potential household respondents. The questionnaire was then administered and data inputted using a computer aided telephone interview system. The final MoNNET-HA sample size was 2707 Montreal residents. Ethics approval for MoNNET-HA was awarded by the Committee of Scientific Evaluation and Research Ethics of the Centre de Recherche at the Centre Hospitalier de l’Université de Montréal (CHUM). Measures Depressive symptoms Depressive symptoms were measured with the Center for Epidemiologic Studies 10-item Depression Scale (CES-D Scale), which can be used to assess current levels of depressive symptoms in adults (Radloff, 1977). Respondents were asked how often they had experienced a series of depressive symptoms in the past week (Bradley, Bagnell, & Brannen, 2010). For example, participants were asked to respond either “yes” or “no” to items such as “I felt happy” or “I felt everything I did was an effort” (Bradley et al., 2010). Based on the recommended cut-off score of four, participants who had replied affirmatively to four or more items were classified as having depressive symptoms (for a list of items see: Irwin, Haydari Artin, & Oxman, 1999). The CES-D scale has been shown to be reliable and valid in the general adult population, as well as in older adults (Irwin et al., 1999). The MoNNET-HA depression scale had a Cronbach’s alpha of 0.72, indicating good reliability.

Core social ties and core tie diversity Core social ties and core tie diversity were measured with the name generator and name interpreter instruments. The name generator asked respondents to list up to three individuals with whom they had discussed important matters within the last 6 months. Respondents who had not discussed important matters with anyone in the past 6 months were grouped into a “no core ties” category. In instances where no core ties were reported, respondents were asked whether this meant that they had no core ties or whether they did not want to respond to the question. Those who did not want to respond were excluded from analyses (n ¼ 72). The core social ties measure compared those with one or more social tie to those with no social ties. Following the name generator, the name interpreter asked respondents whether those they had named resided in their households, neighbourhoods, within the MMA, or outside the MMA. Core tie information was used to create four categories of core tie source: (1) no core ties, (2) neighbourhood ties only (core ties living within the respondent’s neighbourhood), (3) both neighbourhood and non-neighbourhood core ties (i.e., having core ties in the neighbourhood, and either in the household or outside the neighbourhood), and (4) nonneighbourhood ties only (i.e., having household or outside neighbourhood ties). Since our focus was on the relative contribution of neighbourhood ties to depressive symptoms, we used as the reference group those with strictly neighbourhood ties. General and neighbourhood network capital Network social capital was measured with a position generator, which asked respondents if they knew someone on a first-name basis holding specific occupations. If they indicated knowing someone in the occupation, they were then asked if the person lived in their neighbourhood, outside their neighbourhood, within the MMA, or outside the MMA. The position generator consisted of 10 occupations ranging from low to high in occupational prestige (Moore et al., 2011). Occupational prestige scores were used to calculate scores for three dimensions of network social capital: diversity, range, and upper-reachability. Diversity represents the number of different occupations accessed; range is the difference between the highest and lowest prestige occupation accessed; while upper reachability represents the highest prestige occupation in the respondent’s network (Lin, 2001). Factor analysis was used to create a general network social capital score. General network social capital consisted of scores from their social connections, regardless of residence location, whereas neighbourhood network social capital included only those social ties that resided within respondents’ neighbourhoods. Psychosocial and structural measures of social capital

Network social capital A name generator/interpreter instrument sequence and a position generator were used to collect network data in the MoNNETHA study (please refer to Moore et al., 2011 for further descriptions of these instruments). The instruments allow the development of a range of network measures of social capital. To assess the importance of the source of social connections (i.e.,

Generalized trust Generalized trust was measured with the question “Generally speaking, would you say that most people can be trusted or that you can’t be too careful in dealing with people?” Participants selected one of the following options: (1) most people can be trusted, (2) can’t be too careful, (3) depends, (4) most people cannot be trusted,

E. Bassett, S. Moore / Social Science & Medicine 86 (2013) 96e102

and (5) don’t know. Responses were recoded into a dichotomous variable so that those who replied, “most people can be trusted” (i.e., high trust) were contrasted with those who answered affirmatively to one of the other four categories (i.e., low trust). Trust in neighbours A single item measure was used to assess trust in neighbours. Respondents were asked “most people in your neighbourhood can be trusted” and responded based on a five-point Likert scale ranging from strongly disagree to strongly agree. Responses were coded so that a higher score represented higher trust in neighbours. Perceived neighbourhood cohesion Neighbourhood social cohesion was measured by asking respondents to report the extent to which they agreed with the following four statements: (1) you have trouble with your neighbours, (2) people in your neighbourhood are willing to help each other, (3) most people in your neighbourhood know you, and (4) your neighbourhood is clean. A five-point Likert scale ranging from “strongly disagree” (2) to “strongly agree” (2) was used. Item one was reverse-coded and all items summed to create a perceived cohesion score for each participant so that higher scores indicated greater perceived cohesion. Scores were completed only in instances where respondents answered at least three questions. The perceived neighbourhood cohesion scale had a Cronbach’s alpha of 0.32. Social participation Neighbourhood social participation was assessed by asking respondents if they had held a volunteer or officer role in a neighbourhood association or group within the past five years. Asking respondents if they had held a volunteer or officer role in a nonneighbourhood association or group assessed outsideneighbourhood social participation within the past five years. Respondents replied with a “yes” or “no” for each question. Socio-demographic characteristics Several demographic variables were measured and used as covariates in the multilevel analyses. Respondents stated whether they were male or female. Age was measured using six age categories: (1) 25e34 years old, (2) 35e44 years old, (3) 45e54 years old, (4) 55e64 years old, (5) 65e74 years old, and (6) 75 years and older. Marital status was grouped into four categories for these analyses: (1) married or in a common law relationship, (2) single, (3) separated or divorced, and (4) widowed. Education was divided into four categories: (1) no high school certificate or diploma, (2) high school diploma or trade certificate or diploma, (3) college certificate or diploma lower than a bachelor’s degree, and (4) university degree or more. Employment status was measured by asking respondents whether they were currently employed or not. Income was divided into five categories: (1) less than $28 000, (2) $28 000 to $49 000, (3) $50 000 to $74 000, (4) $75 000 to $100 000, and (5) over $100 000. Income data were imputed for 20% of the observations using ordinal regression and information about their educational attainment, employment status, age, and 2006 Canada census data on the median household income of the person’s census tract of residence. Foreign-born status was assessed by asking participants if they were born inside or outside Canada. Respondents stated whether the primary language spoken in the household was French, English, or other.

99

model, socio-demographic variables were first entered to examine whether their association with depressive symptoms. In the second model, general social capital measures were added to the base model. These included: core social ties, general network capital, generalized trust, and social participation. The third model substituted neighbourhood-specific measures for the general social capital measures. These measures included: core tie diversity, neighbourhood network capital, trust in neighbours, neighbourhood social participation, and perceived neighbourhood cohesion. To examine the relative association of general and neighbourhood-specific social capital, the final model included only the social capital variables significant in models two and three. Multilevel logistic regression was used to account for the clustered sampling design in which respondents were nested within census tracts. No area-level variables were included in this study. Odds ratios and 95% confidence intervals are reported. Wald chi-square statistics were used to assess model fit. The response rate of the MoNNET-HA study was 38.7%. The response rate was calculated by dividing the number of completed interviews by the number of incomplete interviews, noninterviews, and estimated proportions of cases of unknown eligibility (Moore et al., 2011). The degree to which the sample is representative of those in the MMA was calculated by comparing the observed sample counts on certain demographic characteristics to expected counts based on the 2006 Canada census. Results showed that the sample over-represented older adults, those with incomes of less than $50 000 per year, those with a residential duration over 5 years, women, and those with college and university levels of education (Moore et al., 2011). Results Sample After excluding observations missing data on study variables, this study had a final sample size of 2624 adults. Of the 2624 participants, 17.3% were classified as having depressive symptoms. Approximately 64.6% of respondents were female, 54.5% of respondents were married, 38.2% had a university degree, 54.8% were employed, 78.1% lived in primarily French speaking households, and 81.5% were born in Canada. More information regarding the socio-demographic variables can be seen in Table 1. Base model: socio-demographic and -economic factors Table 2 presents the results for the base model, model one, and model two. Estimates for the socio-demographic covariates can be found in the Supplementary Table. Among the socio-demographic and -economic factors, gender, age, marital status, and income were each associated with having depressive symptoms. Men were 34% less likely to have high depression scores compared to women (OR ¼ 0.66, 95% CI ¼ 0.52e0.84). Those aged 55e64 were 34% less likely (OR ¼ 0.66, 95% CI ¼ 0.44e0.99), those 65e74 were 50% less likely (OR ¼ 0.50, 95% CI ¼ 0.32e0.78), and those 75 and older were 59% less likely (OR ¼ 0.41, 95% CI ¼ 0.24e0.71) compared to those aged 25e34 to have depressive symptomatology. There was a gradient in risk for depressive symptomatology with those in higher income categories less likely to have depressive symptoms than lower income categories. The Wald chi-square statistic of 108.65 was significant (p < 0.001) indicating that variables within the model provided a good fit to depressive symptoms.

Statistical analyses Model 1: general social capital A four-stage model building process was used to examine if general and neighbourhood psychosocial and network measures of social capital were associated with depressive symptoms. For the base

Among the general measures of social capital, participants with low compared to high levels of generalized trust were 66% more

100

E. Bassett, S. Moore / Social Science & Medicine 86 (2013) 96e102

Variables

Percentage

The Wald chi-square value (157.64, p < 0.001) indicated that model three was significant, and likelihood ratio tests indicated that model 3 provided a better fit than model one (LR chi2 ¼ 39.95, p < 0.001).

Depressive symptomatology Female Age 25e34 years old 35e44 years old 45e54 years old 55e64 years old 65e74 years old 75 þ years old Marital Status Married Single Divorced/Separated Widowed Education No high school High School/Trade Some College University degree or higher Income Less than $28 000 $28 000e49,000 $50 000e74,000 $75 000e100,000 $100 000 and over Employed Foreign born Household language French English Foreign language Social Capital Network measures: No Core Ties Core tie diversity neighbourhood ties only Neighbourhood ties only Non-neighbourhood ties only Mixed locale ties (neighbourhood and non-neighbourhood) Trust and participation measures: High Generalized trust Social participation Neighbourhood participation

17.3 64.6

Model 3: composite (general and neighbourhood social capital)

Table 1 Characteristics of Montreal neighbourhood networks and healthy aging study (MoNNET-HA), social capital and depressive symptoms sample, n ¼ 2624.

Trust in neighbours Perceived neighbourhood cohesion

14.7 17.8 20.2 16.2 20.9 10.3 54.5 20.4 14.9 10.3 11.9 29.3 20.6 38.2 20.4 28.3 26.8 13 11.6 54.8 18.5 78.1 13.7 8.3

13.4 9.2 9.2 44.6 32.9 42.8 36.3 24.0 Mean (SD) 1.12 (0.02) 0.77 (0.01)

likely to report depressive symptoms (OR ¼ 1.66, 95% CI ¼ 1.30e 2.12). Previously significant socio-demographic and economic variables continued to be significant in model two. The Wald chisquare statistic was significant (125.54, p < 0.001) and a likelihood ratio test determined that model two provided a better fit than model one (LR chi2 ¼ 21.78, p < 0.001). Model 2: neighbourhood social capital Among the neighbourhood social capital measures, the psychosocial measures of trust in neighbours and perceived neighbourhood cohesion were associated with having depressive symptomatology. Those with high compared to low levels of trust in their neighbours were 17% less likely to have depressive symptomology (OR ¼ 0.83, 95% CI ¼ 0.76e0.91), while those who perceived their neighbourhoods as being more cohesive were 32% less likely to depressive symptomatology (OR ¼ 0.68, 95% CI ¼ 0.58e0.81). A network measure of social capital, namely core tie diversity, was also associated with depressive symptoms. Those with both neighbourhood and non-neighbourhood ties were 82% more likely to have depressive symptomatology compared to those with neighbourhood-only core ties (OR ¼ 1.82, 95% CI ¼ 1.18e2.82).

Social capital measures that were associated with depressive symptoms in models 2 and 3 were included in the final model. Psychosocial measures remained associated with depressive symptoms with those reporting low generalized trust (OR ¼ 1.46, 95% CI ¼ 1.14e1.88), low trust in neighbours (OR ¼ 0.85, 95% CI ¼ 0.77e0.92), and poorer perceptions of neighbourhood cohesion (OR ¼ 0.71, 95% CI ¼ 0.60e0.84) being more likely to experience depressive symptoms. The network measure assessing “source” core tie diversity continued to be significant with those having neighbourhood-only core ties less likely to report depressive symptoms compared to those with both neighbourhood and non-neighbourhood ties (OR ¼ 1.84, 95% CI ¼ 1.19e2.84). The significant Wald chi-square statistic (160.17, p < 0.001). Discussion The current study examined whether social capital was associated with depressive symptoms in a sample of urban-dwelling Canadian adults. Results indicated that both psychosocial and network measures of social capital were associated with depressive symptoms, and the inclusion of both provides a more comprehensive picture of how social capital is associated with depression. Those with high levels of trust in other people, those who trust their neighbours, and those who perceive their neighbourhoods to have high levels of cohesion were less likely to have depressive symptoms. Our findings are consistent with previous work in finding strong inverse associations between depressive symptoms and intra- and extra-neighbourhood psychosocial measures of trust and social cohesion (De Silva et al., 2005; Echeverria, Diez-Roux, Shea, Borrell, & Jackson, 2008; Kim & Ross, 2009; Mulvaney & Kendrick, 2005; Veenstra, 2005; Whitley & McKenzie, 2005). Stress has been proposed as a potential mechanism linking trust and health, with trusting individuals less likely to experience stress and thus less likely to face negative health outcomes than those who do not trust as easily (Abbott & Freeth, 2008). Neighbourhood research has also suggested stress to be a potential mechanism linking characteristics of the neighbourhood to mental health (Echeverria et al., 2008). Persons living in a more cohesive neighbourhood may have decreased stress and therefore lower rates of depression. Participation measures were not associated with depression in this study, which was consistent with findings by Fujiwara and Kawachi (2008). Differences in associations between social participation, trust, social cohesion, and network measures with depressive symptoms may be due to fundamental differences within these components of social capital. With regards to participants’ core network ties, there were two main findings. First, the study found no association between having or not having core ties, i.e., social isolation, and depressive symptomatology. Previous research has suggested links between social isolation and health (Berkman, 1995; Bruce & Hoff, 1994). For example, Berkman and Syme (1979) showed a link between social isolation and all-cause mortality. Bruce and Hoff (1994) have suggested that social isolation may mediate the relationship between physical health status and depression among a group of adults at greater risk of major depression. Discrepancies in findings between our study and previous work may relate however to the specific population, health outcome, or the specific measure of isolation used. Second, our study showed that certain aspects of people’s core

E. Bassett, S. Moore / Social Science & Medicine 86 (2013) 96e102

101

Table 2 Adjusted odds ratio and 95% confidence intervals of the association between dimensions of social capital and depressive symptomatology, from multilevel logistic regression analyses controlling for socio-demographic and economic variables, n ¼ 2624. Variables Network measures Core ties One or more core tie No core tie Core tie diversity No core ties Non-neighbourhood ties only Neighbourhood and nonneighbourhood ties Neighbourhood ties only General network capital Inside-neighbourhood network capital Trust and participation measures: Generalized trust Low trust High trust Social participation Trust in neighbours Perceived neighbourhood cohesion Neighbourhood participation

Model 1 (General)

Model 2 (Neighbourhood)

Model 3 (Composite)

e

1.19 (0.72e1.98) 1.39 (0.90e2.14)

1.09 (0.66e1.80) 1.30 (0.85e2.00)

e

1.82 (1.18e2.82)**

1.84 (1.19e2.84)**

1.10 (0.97e1.24) e

1.00 e 1.09 (0.97e1.22)

1.00 e e

e

1.21 (0.85e1.71) 1.00

1.66 (1.30e2.12)*** 1.00 1.09 (0.92e1.29) e e

e 0.83 (0.76e0.91)*** 0.68 (0.58e0.81)***

1.46 (1.14e1.88)** 1.00 e 0.85 (0.77e0.92)*** 0.71 (0.60e0.84)***

e

1.16 (0.89e1.52)

e

*p < 0.05, **p < 0.01, ***p < 0.001.

network ties were associated with depressive symptoms. Specifically, individuals who had only neighbours as core ties compared to those with both neighbourhood and non-neighbourhood core ties were less likely to have depressive symptomatology. Features of neighbourhood environments are increasingly considered as important in studies of depression, and knowing one’s neighbours has been associated with decreased depression in previous studies (Caughy, O’Campo, & Muntaner, 2003; Mair, Diez Roux, & Galea, 2008; Mulvaney & Kendrick, 2005). Yet, in the current study, no differences were found between those whose core ties resided exclusively within the neighbourhood compared to those whose core ties lived exclusively outside of neighbourhood boundaries. Instead, those having core ties spanning neighbourhood and nonneighbourhood locations were at greater odds of having depressive symptoms compared to those with neighbourhood ties only. We are not aware of any previous research to investigate the degree to which mixed source ties lead to different mental health outcomes. Future research might investigate the specific processes by which neighbourhood only core ties are protective against depressive symptoms compared to mixed source locations. Network social capital was not shown associated with depressive symptomatology in the current study. Webber et al. (2010) also found depression unrelated to network capital in their study (Webber et al., 2010). The lack of an association between network social capital and depressive symptomatology may be due to the type of network ties that a position generator appears to capture. Previous research using position generators to measure social capital have suggested that position generators tend to capture people’s weaker social ties (Lin, 2001; Moore, Daniel, Gauvin, et al., 2009; Moore, Daniel, Paquet, et al., 2009). While weak ties may play a greater role in shaping health behaviour, e.g., physical inactivity (Legh-Jones & Moore, 2012), they may not provide the type of support or resources that would help reduce the chances of depression. Limitations The current study has some unavoidable limitations. Due to the cross-sectional design, cause-effect relationships cannot be

established. For example, it is unknown whether having low social capital results in depression, or whether having depression causes individuals to become disconnected within their networks and decrease their use of resources thus resulting in lower social capital. Previous research has shown complex reciprocal effects when it comes to the relationship between depression and social relationships (Chou & Chi, 2003; Patten, Williams, Lavorato, & Bulloch, 2010; Ramos & Wilmoth, 2003). Our study takes a sociological approach and focuses on the importance of social factors on individual mental health. A second limitation to this study is that depressive symptoms are not clinician-rated. Although the CES-D is a valid and reliable screening tool for measuring depressive symptoms, it is not able to make clinical diagnoses of depression (Radloff, 1977). As a result, conclusions in this study are limited to depressive symptoms, but not major depressive disorder itself. Lastly, our perceived cohesion scale had a low reliability. To confirm findings, we constructed a perceived cohesion factor score using principal components analysis. In ancillary analyses, we substituted the cohesion factor for the scale. Results were similar to those using the scale. Conclusions Cross-sectional associations between social capital and depressive symptoms are generally well established. Yet, there is a need to understand better the specific mechanisms by which social capital is associated with depressive symptoms. Recognizing that the association between social capital and depression may differ according to the specific dimension of social capital assessed is a critical step in identifying the specific mechanisms by which social capital is associated with health, and in designing effective interventions aiming to decrease depression. Acknowledgements This study was funded by an operating grant from the Canadian Institutes of Health Research (MOP 84584). At the time of the research and analysis, SM held a New Investigator Award from the Canadian Institutes of Health Research e Institute of Ageing

102

E. Bassett, S. Moore / Social Science & Medicine 86 (2013) 96e102

Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.socscimed.2013.03.005. References Abbott, S., & Freeth, D. (2008). Social capital and health: starting to make sense of the role of generalized trust and reciprocity. Journal of Health Psychology, 13(7), 874e883. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, D.C.: Author. Aslund, C., Starrin, B., & Nilsson, K. W. (2010). Social capital in relation to depression, muskoskeletal pain, and psychosomatic symptoms: a cross-sectional study of a large population-based cohort of Swedish adolescents. BMC Public Health, 10, 715e725. Berkman, L. F. (1995). The role of social relations in health promotion. Psychosomatic Medicine, 57, 245e254. Berkman, L. F., & Syme, L. (1979). Social networks, host resistance, and mortality: a nine-year follow-up study of Alameda County residents. American Journal of Epidemiology, 109(2), 186e204. Bourdieu, P. (1986). Forms of capital. In J. G. Richardson (Ed.), Handbook of theory for the sociology of education (pp. 241e258). Westport, CT: Greenwood Press. Bradley, K. L., Bagnell, A. L., & Brannen, C. L. (2010). Factorial validity of the center for epidemiological studies depression 20 in adolescents. Issues in Mental Health Nursing, 31, 408e412. Bruce, M. L., & Hoff, R. A. (1994). Social and physical health risk factors for firstonset major depression disorder in a community sample. Social Psychiatry and Psychiatric Epidemiology, 29, 165e171. Carpiano, R. M. (2006). Toward a neighbourhood resource-based theory of social capital for health: can Bourdieu and sociology help? Social Science & Medicine, 62, 165e175. Carpiano, R. M., & Hystad, P. W. (2011). “Sense of community belonging” in health surveys: what is it measuring? Health & Place, 17, 606e617. Caughy, M. O., O’Campo, P. J., & Muntaner, C. (2003). When being alone might be better: neighborhood poverty, social capital, and child mental health. Social Science & Medicine, 57, 227e237. Chou, K. L., & Chi, I. (2003). Reciprocal relationship between social support and depressive symptoms among Chinese elderly. Aging & Mental Health, 7(3), 224e231. De Silva, M. J., McKenzie, K., Harpham, T., & Huttly, S. R. (2005). Social capital and mental illness: a systematic review. Journal of Epidemiology & Community Health, 59, 619e627. Echeverria, S., Diez-Roux, A. V., Shea, S., Borrell, L. N., & Jackson, S. (2008). Associations of neighborhood problems and neighborhood social cohesion with mental health and health behaviors: the Multi-Ethnic Study of Atherosclerosis. Health & Place, 14, 853e865. Fujiwara, T., & Kawachi, I. (2008). A prospective study of individual-level social capital and major depression in the United States. Journal of Epidemiology & Community Health, 62, 627e633. Haines, V. A., Beggs, J. J., & Hurlbert, J. S. (2011). Neighborhood disadvantage, network social capital, and depressive symptoms. Journal of Health and Social Behavior, 52(1), 58e73.

Irwin, M., Haydari Artin, K., & Oxman, M. N. (1999). Screening for depression in the older adult: criterion validity of the 10-item center for epidemiological studies depression scale (CES-D). Archives of Internal Medicine, 159, 1701e 1704. Kim, J., & Ross, C. E. (2009). Neighborhood-specific and general social support: which buffers the effect of neighborhood disorder on depression? Journal of Community Psychology, 37(6), 725e736. Legh-Jones, H., & Moore, S. (2012). Network social capital, social participation, and physical inactivity in an urban adult population. Social Science & Medicine, 74, 1362e1367. Lin, N. (1999). Building a network theory of social capital. Connections, 22(1), 28e51. Lin, N. (2001). Social capital: A theory of social structure and action. Cambridge: Cambridge University Press. Mair, C., Diez Roux, A. V., & Galea, S. (2008). Are neighbourhood characteristics associated with depressive symptoms? A review of evidence. Journal of Epidemiology & Community Health, 62, 940e946. Moore, S., Daniel, M., Gauvin, L., & Dubé, L. (2009). Not all social capital is good capital. Health & Place, 15, 1071e1077. Moore, S., Bockenholt, U., Daniel, M., Frohlich, K., Kestens, Y., & Richard, L. (2011). Social capital and core network ties: a validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health. Health & Place, 17(2), 536e544. Moore, S., Daniel, M., Paquet, C., Dubé, L., & Gauvin, L. (2009). Association of individual network social capital with abdominal adiposity, overweight and obesity. Journal of Public Health, 31(1), 175e183. Moussavi, S., Chatterji, S., Verdes, E., Tandon, A., Patel, V., & Ustun, B. (2007). Depression, chronic diseases, and decrements in health: results from the World Health Surveys. The Lancet, 370, 851e858. Mulvaney, C., & Kendrick, D. (2005). Depressive symptoms in mothers of pre-school children effects of deprivation, social support, stress and neighbourhood social capital. Social Psychiatry and Psychiatric Epidemiology, 40, 202e208. Patten, S. B., Williams, J. V. A., Lavorato, D. H., & Bulloch, A. G. M. (2010). Reciprocal effects of social support in major depression epidemiology. Clinical Practice & Epidemiology in Mental Health, 6, 126e131. Radloff, L. S. (1977). The CES-D scale: a self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385. Ramos, M., & Wilmoth, J. (2003). Social relationships and depressive symptoms among older adults in southern Brazil. The Journals of Gerontology: Series B, 58(4), 253e261. Rosenquist, J. N., Fowler, J. H., & Christakis, N. A. (2011). Social network determinants of depression. Molecular Psychiatry, 16, 273e281. Sund, E. R., Jorgensen, S. H., Jones, A., Krokstad, S., & Heggdal, M. (2007). The influence of social capital on self-rated health and depression e the Nord-Trondelag health study (HUNT). Norwegian Journal of Epidemiology, 17(1), 59e69. Veenstra, G. (2005). Location, location, location: contextual and compositional health effects of social capital in British Columbia, Canada. Social Science & Medicine, 60, 2059e2071. Webber, M., Huxley, P., & Harris, T. (2010). Social capital and the course of depression: six-month prospective cohort study. Journal of Affective Disorders, 129, 149e157. Whitley, R., & McKenzie, K. (2005). Social capital and psychiatry: review of the literature. Harvard Review of Psychiatry, 13(2), 71e84.